Membrane oxygenator longevity was higher in argatroban-treated patients undergoing vvECMO

Background: In severe acute respiratory distress syndrome (ARDS),


| BACKGROUND
Extracorporeal membrane oxygenation (ECMO) is a lifesupporting technique that can be used in patients with respiratory failure. In the ECMO cycle, blood interacts with large nonendothelial areas of foreign surfaces, activating coagulation cascades. This increases the risk of clot formation within the venous system or in the membrane oxygenator. To reduce the risk of clotting, patients with ECMO are treated with anticoagulant medications. However, these medications increase the risk of bleeding.
The most used anticoagulant in vvECMO is heparin. Heparin was first discovered in 1916 and was introduced to the clinic shortly afterwards in the 1920s. 1 Despite its long use, heparin therapy has important risks: The most serious adverse effect of heparin is the development of heparininduced type II thrombocytopenia (HIT II). Although HITII is relatively rare in the critically ill, affecting approximately 1% of patients, 2,3 it can have disastrous consequences if the diagnosis is missed. Among patients with serologically confirmed HIT II, 50% to 70% will develop venous or arterial thrombosis. 4 In other cases, clinicians observed a phenomenon called heparin resistance, defined as the need for high doses of heparin to achieve the target level of anticoagulation. 5 In the perioperative setting, approximately 8% of patients are classified as resistant to heparin. 6 Newer anticoagulants, such as direct thrombin inhibitors (DTI), offer the advantage that they can bind directly to circulating or clot-bound thrombin without the need for a cofactor. 7 When there are contraindications to heparin, such as in HITII, direct thrombin inhibitors can be used for prophylaxis against thromboembolism. Several agents are available for continuous intravenous use, with bivalirudin and argatroban as the main agents. A recent meta-analysis found that argatroban outperforms other DTI in patients with HIT II with the shortest hospitalization and the lowest rate of haemorrhage, thromboembolism and mortality. 8 Argatroban has a short half-life of about 15 min and a fast metabolism. These are benefits for patients with ECMO support, although there is no specific antagonist for argatroban.
We hypothesized that argatroban induces stronger inhibition of coagulation, leading to a longer operating time of individual membrane oxygenators in patients with vvECMO. To assess this, we identified patients who underwent vvECMO for severe acute respiratory distress syndrome. Furthermore, we measure whether argatroban impacts the need for transfusion of blood products as a surrogate for bleeding complications.

| Study population and ethics
All data was retrospectively collected from medical records from the University Hospital of Tuebingen. The study was approved by the Ethics Committee of the University Hospital Tuebingen (IRB# 250/2021BO2), which waived the need for informed consent because patient anonymity was maintained.

| Study subjects
All patient records that were treated at the Department of Anesthesiology and Intensive Care Medicine between January 2019 and February 2021 were retrospectively screened for inclusion criteria. We identified 65 patients who underwent vvECMO during the study period; 8 patients had to be excluded due to initial vaECMO cannulation, which we later changed to vvECMO due to secondary respiratory failure. The main inclusion criterium was vvECMO treatment for severe acute respiratory distress syndrome. Patients who received ECMO with different cannulations as venovenous (e.g., venoarterial) were excluded from the analysis. We identified 65 patients who underwent vvECMO. Eight patients did not meet the inclusion criteria and were subsequently excluded from the analysis. Among the remaining 57 patients, 33 received unfractionated heparin as anticoagulation and 24 patients received argatroban.
Based on data from the clinical information system, a database was generated that contains relevant patient information, including age, sex, date of admission and discharge from the ICU, and Sequential Organ Failure Assessment (SOFA), Simplified Acute Physiology Score II (SAPS II).

| Anticoagulation strategy during ECMO support
Before 2020, heparin was used according to our institutional standard anticoagulation for patients receiving vvECMO support. With the outbreak of the COVID-19 pandemic, the first-line anticoagulation strategy was changed to Argatroban due to pathophysiological considerations (see the Results section for details). Upon

K E Y W O R D S
acute respiratory distress syndrome, anticoagulation, argatroban, COVID-19, extracorporeal membrane oxygenation implantation of cannulas for vvECMO patients received a fixed bolus injection of 5000 IU of unfractionated heparin. After establishment of an extracorporeal circuit, an activated partial thromboplastin (aPTT) was measured. When aPTT was within target range, continuous administration of anticoagulant was initiated. The target activated partial thromboplastin time (aPTT) was 50-60 s in both groups.

| Transfusion protocol during vvECMO
Criteria for transfusion of packed red blood cells (RBC), platelets and fresh frozen plasma during vvECMO were based on the national guidelines of the German Medical Association (Bundesärztekammer). In summary, transfusions for RBC was a haemoglobin level below 7 mg/dL in the point-of-care diagnostic. Between a haemoglobin concentration between 7 and 9.9 mg/dL, RBC transfusions were performed when indicators of impaired compensation were present. When the haemoglobin concentration exceeded 10 mg/dL, no RBC was performed.
During vvECMO, platelets were substituted when the platelet count was below <50,000/μl. When bleeding complications were present, platelets transfusions were performed even at higher platelet counts at the discretion of the attending physician. Fresh frozen plasma was only transfused in the setting of serve haemorrhage according to national guidelines. Severe haemorrhage was defined as blood loss causing hemodynamic instability.

| Clinical criteria for vvECMO oxygenator exchange
The final decision to exchange the oxygenator was based on a multidisciplinary discussion of the care team. A certified clinical perfusionist assessed oxygenator functionality daily. The following parameters with cut-off values were evaluated:

| Study outcomes
The primary end point was days per oxygenator. This was calculated based on the number of oxygenators used per patient per treatment period. Secondary endpoints were difference in the transfusion volumes of packed red blood cells, thrombocytes and fresh frozen plasma. To standardize between different patients, we normalized these volumes to days of vvECMO treatment.
Bleeding complications were determined based on a review of the patient's chart, if mentioned in the discharge note of the patients.

| Statistical analysis
To reduce bias only complete dataset analysis was performed. All variables were tested for normal distribution using the Shapiro-Wilk goodness-of-fit test. Normal distributed variables are expressed as the mean ± standard and as the median (interquartile range) in the case of a nonnormal distribution. Normally distributed variables were compared using the pooled Student's t-test, and nonnormally compared with the Mann-Whitney U test. Differences between the two study groups were evaluated using the Chisquare independence test for nominal variables. When one group had fewer than five occurrences of a categorical variable, Fisher's exact test was used. p-values are two-tailed and values <.05 are considered statistically significant. For a post hoc power analysis for the primary outcome (oxygenator runtime per vvECMO treatment duration) we sued the following setting: alpha 0.1, standard-deviation heparin group 4.9, argatroban group 9,93. To assess the association of COVID-19 with the duration of the vvECMO support, a univariate nominal logistic regression was used and the area under the curve (AUC) of the receiver operating curve (ROC) was calculated. All statistical and power analysis for this study was performed on Prism 9 (GraphPad Software Inc.) and JMP 16.0 (SAS Institute Inc.).

| Demographic data and patient characteristics
The selection process is depicted in Figure 1. We reviewed the patient records of all patients who received vvECMO for severe acute respiratory distress syndrome during January 2019 and February 2021. We eliminated eight patients from the analysis because they received vaECMO circulatory support before or after vvECMO. Next, we screened the anticoagulation strategy in each patient. The standard of care in our institution at the beginning of the investigation period was an anticoagulatory regime based on unfractionated heparin. At the beginning of the COVID-19 pandemic, we switched to argatroban-based anticoagulation, due to better inhibition of thrombin. This strategic decision was based on the observation that patients with COVID-19 are in a prothrombotic state.
In our dataset, 58% (n = 33) of the patients received heparin as a primary anticoagulation strategy, while 42% (n = 24) received Argatroban for therapeutic anticoagulation. Baseline demographic characteristics, such as age, sex, height, weight and body mass index, were similar between the groups ( Table 1). Note that ICU severity scores, such as the SAPSII or SOFA score, did not differ between the groups. The primary ARDS rate was similar between both groups, but there were significantly more patients Age with ARDS associated with COVID-19 in the argatroban group than those treated with heparin (p = .0002).
In summary, demographic parameters, ICU severity indices and comorbidities did not differ between the groups. There were significantly more patients with COVID-19 in the argatroban-treated group.

| vvECMO treatment variables
Next, we investigated whether the vvECMO treatment variables differed between the two groups. Similar vvECMO systems were used in the different groups. However, the ECMO runtime was significantly different between patients treated with argatroban and heparin ( Table 2). Since mainly patients with COVID-19 received argatroban as an anticoagulant, we performed a univariate logistic regression to test whether COVID-19 was associated with a longer duration of ECMO support. We found that COVID-19 was indeed associated with a longer duration of ECMO (ROC-AUC 0.67804; p < .0250). Although argatroban-treated patients had longer vvECMO treatment times, the number of membrane oxygenators required was the same between the two groups. Furthermore, the lowest platelet count and the course of platelet numbers in the first 6 days after starting vvECMO treatment did not differ (Figure 2A). Prothrombin time was significantly higher in the argatroban group (argatroban: 53 ± 7 s vs. heparin 49 ± 8 s; p = .0467; Table 2). Patients treated with heparin. As expected, the international normalized ratio (INR) was higher in the argatroban treatment group, which is a known effect.

| Absolute transfusions in patients treated with argatroban and heparin
Next, we assessed whether the anticoagulation regimen influences the risk of haemorrhage. As an objective surrogate parameter, we quantified the different blood products transfused in our patients. Most of the patients required red blood cell transfusions during their stay in the ICU (argatroban n = 23, 96%; heparin n = 28; 85%). The decision for transfusion was based on our institutional guidelines (depicted in Figure 3). In contrast, only 21% of patients required platelet application, regardless of the group. The number of individual coagulation factors did not differ between the two groups ( Table 3). The number of patients receiving substitution with Antithrombin III (ATIII) was higher in the heparin group. However, the total amount of ATIII substitution did not differ between the two treatment groups (Table 3). Additionally, the mean ATIII serum activity did not differ between the groups, even though the argatroban group trended towards higher values ( Figure 4).

| Differences in bleeding complications between heparin-or argatroban-treated patients
After we found no differences in transfusion requirements between groups, we next assessed if anticoagulation regimes influenced bleeding complications during vvECMO. Bleeding complications were common in our study cohort (40% of all patients experienced at least one bleeding complication). However, bleeding complications did not differ with anticoagulation with heparin or argatroban. Therefore, in our cohort, the anticoagulation regime did not appear to influence the rate of bleeding complications (Table 4).

| Primary and secondary endpoints: membrane oxygenator runtime and transfusion requirements normalized to the duration of vvECMO therapy
As a primary endpoint, we investigated whether argatroban anticoagulation increased the runtime of each oxygenator compared to heparin anticoagulation. To do this, we determined the number of oxygenators for each patient and normalized this number to the duration of vvECMO therapy. As shown in Table 5, we found that argatroban treatment increases the longevity of the individual oxygenator from a mean of days 12.3 (heparin group) to 16.6 days (argatroban group; p = .0365; power = .6186).
Since argatroban can increase bleeding complications in critically ill adults, as a secondary endpoint, we investigated the volume of transfusion products normalized on the day of vvECMO support. We found that in patients who received a transfusion of the respective blood products, the volume of platelets and fresh frozen plasma was increased. Therefore, in patients with argatroban anticoagulation, the net requirement for platelets or fresh frozen plasma was reduced during vvECMO treatment, while the volume of erythrocytes was the same between the groups.
In summary, we found that patients with an argatrobanbased anticoagulatory regimen had a longer duration of oxygenator and required less volume of platelets or fresh frozen plasma transfusion.

| DISCUSSION
In this retrospective cohort study, we compared anticoagulatory regimens in patients with vvECMO. Our primary endpoint was the difference in membrane oxygenator runtime and transfusion requirements of packed erythrocytes, platelets, or fresh frozen plasma per day in vvECMO. We found that patients anticoagulated with argatroban versus heparin had a longer oxygenator runtime when normalized to the overall duration of treatment. Additionally, they received fewer platelet transfusions or fresh frozen plasma transfusions. Our findings suggest that argatroban anticoagulation is feasible in vvECMO patients and is associated with an increase in the longevity of membrane oxygenators. Argatroban does not increase the overall risk of haemorrhage compared to heparin.
During vvECMO treatment, blood is exposed to a large foreign surface, triggering activation of the coagulation cascade and subsequent formation of venous thrombose in the vvECMO circuit. The interaction of blood and the surface of the ECMO circuits can lead to a hypercoagulable state, which can lead to thrombosis. Thus, anticoagulation and vvECMO therapy are an essential condition. Systemic anticoagulation is the only way to prevent cannula, oxygenator, or circuit thrombosis. However, anticoagulation needs to be balanced to reduce the risk of bleeding in patients. This can occur in all patients. In particular, the von-Willebrand factor is depleted from the circulation, which can lead to acquired von-Willbrand-Juergens syndrome. 9,10 As a result, constant low-level thrombotic activation depletes coagulation factors and exposes patients to bleeding complications. As a result, thrombosis and bleeding at the same time are common complications for patients treated with vvECMO. In fact, approximately 45% of patients develop haemorrhage during vvECMO, 11,12 some of them severe. Simultaneously, vvECMO presents a strong procoagulatory stimulus due to the large foreign surface to which the blood is exposed; in fact, thrombotic events such as deep venous thrombosis are more common than bleeding events in vvECMO. 12,13 The membrane oxygenator is particularly at risk of clotting-induced obstruction, because of its larger surface area. This poses a serious health risk to patients undergoing vvECMO treatment because they often completely depend on the membrane oxygenator for respiratory gas exchange. Therefore, according to international guidelines, patients undergoing vvECMO treatment should receive prophylaxis against thrombotic disease, although these guidelines do not specify the agent that should be used for anticoagulation during vvECMO. Unsurprisingly, heparin remains the most widely used anticoagulant. 14 Here, a high-dose heparin strategy appears to be superior to a low-dose heparin strategy 15 to reduce the risk of venous thrombosis. Based on the aPTT measured, we have slightly lower heparin action in the heparin group of our study cohort. This might have contributed to the difference between groups and the oxygenator runtime at first glance. Nevertheless, we are confident that our finding represents an effect of argatroban rather than a low heparin dose. The main concern is an immediate thrombosis of the cannula or oxygenator after implantation of the vvECMO cannula. Thus, all patients received a bolus of heparin regardless of the subsequent anticoagulatory regimen during vvECMO therapy (unless heparin-induced thrombocytopenia is suspected/ known). Therefore, virtually all patients are therapeutically anticoagulated with aPTT >120 s with vvECMO cannula implantation. Repeated coagulation analysis is used to determine the dosing of heparin or argatroban.

F I G U R E 3
Institutional transfusion standard-operating procedure for red blood cells. Therefore, we interpret the difference and the less robust aPTT value in the heparin as an effect of the specific anticoagulatory agent. This is in line with other studies that have found that argatroban might provide a superior and more reliable inhibition of circuit thrombosis. 16 In the other setting, argatroban has become an alternative for critically ill patients. Argatroban exerts anticoagulant activity through direct reversible inhibition of soluble thrombin and clot-bound thrombin. In critically ill patients, argatroban has been used as an alternative anticoagulant, when heparin cannot be used. It has a comparable risk profile, 17 however, argatroban is still much less used for thromboprophylaxis during vvECMO. 14 In most cases, patients received argatroban in the setting of heparin-induced thrombocytopenia, 18 but it might be a safe alternative even without HIT II. In fact, a small prospective clinical trial found that argatroban achieved the PTT target faster than heparin and, therefore, could offer some advantages over heparin in this setting. 19 In our study, we found that patients treated with argatroban for thromboembolic prophylaxis needed fewer blood product transfusions compared with patients receiving heparin. It is interesting to see that even in the setting where PTT was comparable between both study cohorts, transfusion requirements were reduced. This is consistent with other studies, which also found a lower transfusion rate when Argatroban was used as an anticoagulant. 20 Bleeding remains a major factor limiting the outcome of patients with vvECMO, 21 and in early studies with argatroban as an anticoagulant for ECMO. However, when some dosing conditions are considered, clinicians can use argatroban safely in patients with vvECMO. 16,22,23 When focusing on platelet transfusion, we and others found that  argatroban treatment reduces the requirement for platelet transfusion. 20 Platelet counts in the first 6 days after ECMO initiation were not different between the groups.
In our study, we found that the longevity of the membrane oxygenator increased when argatroban was used as the primary anticoagulant. This finding was rather surprising to us, as most of the patients treated with argatroban were admitted to our institution due to ARDS due to COVID-19. Other studies found that COVID-19 in general increases thromboembolic events in general, 24 as well as in patients with vvECMO. 25 Interestingly, this observation could point in the direction that a direct thrombin inhibitor leads to slower obstruction of the membrane oxygenator. For example, in a retrospective analysis of 33 patients with COVID-19 who underwent ECMO, only 1 (6.1%) developed thrombotic events, 26 when the direct thrombin inhibitor bivalirudin was used. Unfortunately, our study is too small to examine whether direct thrombin inhibitors also reduce mortality in this population of patients.

| LIMITATIONS
Here, we report a single-centre retrospective analysis, which is the inherent weakness of limited generalizability. Furthermore, we can only report an association, due to our study design, a direct correlation of our findings, and the dosing of argatroban is limited. Furthermore, in our argatroban group, COVID-19-induced ARDS was overrepresented, which could also have influenced the F I G U R E 4 Mean Antithrombin III serum activity during vvECMO therapy. Antithrombin III (ATIII) serum levels were measured at least twice a week during therapy regardless of anticoagulatory agent at different times point during vvECMO therapy. Laboratory values were retrieved from the medical records and mean levels of ATIII activity calculated for each patient and plotted here. (Argatroban n = 24; Heparin n = 33; compared by Mann-Whitney test; p = .1798). Intracranial haemorrhage 6 (11) 3 (13) 3 (9) .171

Total
Gastrointestinal bleeding 12 (21) 6 (25) 6 (18) .5330 Epistaxis 8 (14) 4 (17) 4 (12) .6257 Pulmonary bleeding 5 (9) 1 (4) 4 (12) .2946 Puncture sites 10 (18) 6 (25) 4 (12) .2069 Other haemorrhage 6 (11) 3 (13) 3 (9) .6788 T A B L E 4 Bleeding complications. overall study cohort. This could affect our results, since COVID-19 is a disease that heavily interferes with coagulation. Nevertheless, we assume that this is a rather supporting fact for our conclusion since COVID-19 increases the ECMO-complications and subsequently the requirement for circuit change. This notion is based on the observation that COVID-19 rather shifts the coagulation cascade in the direction of thrombotic complications. We cannot also rule out the possibility that the management of patients with ARDS and transfusion triggers changed with time. We addressed this limitation by limiting our investigation period to 2 years. Finally, we only have 57 patients in our analysis, so a small effect and difference between the heparin and argatroban groups could have been lost.

| CONCLUSION
In summary, we report that argatroban as an anticoagulant in vvECMO patients increases the longevity of the vvECMO circuit. Furthermore, argatroban as the primary anticoagulant reduces the transfusion requirement when normalized to the vvECMO runtime compared to heparin. Our findings are an independent confirmation of other studies, underscoring the urgency of a prospective randomized controlled trial.